Topological Approach to Analgesia - American Chemical Society

Apr 4, 1994 - assisted design of new drugs, combining them with other indices of our own.5-7 In this study, we apply different topological indices inc...
0 downloads 0 Views 585KB Size
J . Chem. Znf. Comput. Sci. 1994, 34, 1198-1203

1198

Topological Approach to Analgesia J. GQlvez,’-+R. Garcia-Domenech, V. De Juliln-Ortiz, and R. Soler Unit Research into Molecular Connectivity and Drug Design, Department of Physical Chemistry, Faculty of Pharmacy, University of Valencia, Valencia, Spain Received April 4, 1994” In this study we show that by making use of the molecular connectivity indices, including a new index, which we denominate ‘‘a? and which is obtained from a linear combination of the indices, as well as from an “E” form index, it is possible to discriminate minor (nonnarcotic) analgesic character with great efficiency, as well as forseeing the analgesic potency of the analyzed compounds, which represents, without a doubt, a powerful tool for the rational design of new analgesic drugs. 1. INTRODUCTION

Connectivity indices have shown their usefulness in the prediction of diverse physical, chemical, and biological properties of an extensive group of compounds.14 Our research team has applied connectivity indices to thecomputer assisted design of new drugs, combining them with other indices of our In this study, we apply different topological indices including connectivity as well as others derived from them, for the analgesiccharacter discrimination of an extensive group of compounds, as well as for the prediction of its relative potency if the considered compound is analgesic. The format of this paper is as follows. Definitions and an outline of our procedures are described in section 2. The obtained results are described and discussed in section 3. Finally, the practical conclusions of the work are summarized in the last section.

With the objective of reaching a discriminant function of the analgesic character using connectivity indices, we proceeded with the selection of groups of simple molecules (phenyl derivatives), with the greatest resemblance (including isomeric position), one being analgesic and the other/s not. All of these compounds are derived from the basic structure. R’

I

&R,R

On the one hand, different structural and physicochemical parameters, among which we can emphasize the charge on atoms, dihedral angles, interatomic distances, formation enthalpy, ionization potentials, and the dipolar moments, were calculated by mechanoquantical procedures (AM 1 method).* Of all of these properties, only the last, i.e. the dipolar moment, showed that it kept some kind of relation, as we shall see further on. On the other hand, with the objective of evaluating the inductive and mesomeric effects, which are clearly related to the properties of electrons “T” and “nn, the differences between nonvalence and valence indices (’x - lxV,*x - 2xv, etc.) were calculated, since these differences seem to evaluate said proper tie^.^ From these differences, the most significant ones were 4xp- 4xvpand 4xpc- 4xvpc. t Current address: c/Ciscar, 40, 46005 Valencia, Spain.

Abstract published in Aduance ACS

Abstracts.

compound

activity

p-methylpropiophenone p-chloropropiophenone

yeslo

noLo

2,4-dimethylacetophenone yeslo 3,4-dimethylacetophenone nolo 2 4 1-propenyl)phenol 2-(2-propenyl)phenol o-hydroxybenzoic acid m-hydroxybenzoic acid p-hydroxybenzoic acid o-hydroxybenzamide m-hydroxybenzamide p -hydroxybenzamide 3-chlorosalicyclic acid 5-chlorosalicyclic acid

yesLL noLZ yesL4 noL4 n014

yesL4 noI4 nol4 yeslo noLo

A4xP p (D) 0.050 0.016 0.084 0.046 0.1453 0.1557 0.397 0.271 0.234 0.370 0.251 0.212 0.335 0.328

2.601 2.164 2.602 2.704 0.96 1.140 1.474 1.373 1.400 2.920 3.980 3.270 2.378 1.371

A4xP

ut

0.139 0.099 0.102 0.1 11 0.1843 0.2050 0.375 0.321 0.329 0.351 0.295 0.303 0.450 0.338

-0.049 0.043 0.022 -0.020 0.0011 -0.009 0.053 -0.010 -0.054 0.058 -0.004 -0.050 -0.075 0.030

Table 2. Potency, P, (Relative to ASA = l), IC50,E, ut, and log IC50 Values of a Group of Analgesics

2. METHODOLOGY

@

Table 1. Values of Different Physicochemical and Topological Parameters for Sets of Analogous Compounds with and without Analgesic Activity

July 15, 1994

compound

P

1C5ob(pM)

ASA salicyclic acid flufenamic acid meclofenamic acid mefenamic acid alclofenac aminopyrine diclofenac etodolac fenclofenac phenylbutazone fenoprofen flurbiprofen acetaminophen indomethacin ketoprofen naproxen piroxicam tolmetin sulindac ibuprofen

1 0.8

700 1000 6 15 4 18 10000 4.6 169 100 0.34 10.5 6 32 120 11.7 28 45.2

1

1 2 1.5 1.5 40 1.5 1 0.7 3.0 20 1.2 40 30 10 50 15 40 28

UI

0.1682 0.0626 0.443 0.139 0.144 0.2146 0.372 0.109 0.77 0.016 0.401 0.0028 -0.109 0.109 0.242 -0.207 -0.107 -0.415 0.165 0.086 -0.170

E log ICso‘ 2.15 2.97 2.46 2.83 1.59 0.53 1.89 0.57 1.86 1.16 0.90 1.71 1.34 3.68 1.57 0.80 1.81 1.69 1.39 1.52 1.45 1.88 1.29 1.34 1.25 1.21 1.42 2.45 1.08 1.08 1.33 0.68 1.29 1.01 1.44 2.54 0.82 1.44 1.08 0.95 1.13 1.24

Data obtained from refs 13 and 15. Inhibition concentration-50

(ICso)in bovine seminal vesicle (data obtained from ref 13). Resulting values from eq 5.

3. RESULTS AND DISCUSSION

Table 1 shows the values of said differences together with the theoretically determined dipolar moments for the chosen molecular groups. Several facts stand out: On the one hand, the difference 4xp- 4xpv, A4xpseems to be more sensitive to the modification

0095-2338/94/1634-1198S04.50/0 0 1994 American Chemical Society

J. Chem. Znf. Comput. Sci., Vol. 34, No. 5, 1994 1199

TOPOLOGICAL APPROACH TO ANALGESIA

Table 3. Results Obtained by the Discriminant Analysis, Carried Out on a Group of 82 Molecules (40 with Analgesic Activity (+) 42 Inactives

(-)la compounds with analgesic activity molecule

Yl

AAS acetanilide flufenamic acid alclofenac meclofenamic acid mefenamic acid aminopyrine 4-hydroxyisophthalic aminopropylon benorylate benoxaprofen benzpiperylon carprofen clopirac diclofenac diflunisal droxicam phenylbutazone fenoprofen feprazone flurbiprofen ibuprofen idomethacin isonixin isoxicam ketoprofen ketoprolac naproxen niflumic acid oxaprozin oxyphenbutazone acetaminophen piroxacam propylphenazone salicyclic acid salsalate suiindac tenoxicam tiaprofenic tolmetin

0.42 0.41 0.47 0.20 0.42 0.47 0.42 0.32 -0.72 0.46 -0.04 0.3 0.23 0.46 0.43 0.22 0.47 0.47 0.42 0.41 0.37 0.46 0.47 0.47 0.37 0.40 0.45 0.36 0.46 0.47 0.28 0.42 0.46 0.35 0.47 0.47 0.44 0.44 0.47 0.40

log IC 2.97 2.26 0.53 1.71 0.57 1.16 3.68 3.00 3.00 2.47 1.05 0.84 0.91 1.60 0.80 2.73 3.22 1.88 1.34 1.22 1.21 1.24 1.08 1.89 3.37 0.68 2.07 1.01 0.89 1.23 2.08 2.45 2.54 3.14 2.83 2.59 0.95 1.82 -0.44 1.44

compounds without analgesic activity classification

+ + + + + + + ++ + + + ++ + + + + + ++ + + + + + + ++ + + + +

molecule aciclovir desciclovir ebastatine acenocoumarol afloqualone amaranth (dye) ascorbic acid benziodarone bromindione bromazine carisoprodol a-carotene @-carotene y-carotene carteolol cloranolol erythritol histapyrodine isopromethazine khellin oxazidione tricromyl bamipine ursolic acid carnitine EPA oxatomine promethazine terfen vidarabine rimantadine AzddU AzddCF3U AzddFU AzddIU AzddOEtU tertatolol timolol &tocopherol toliprolol triamterene xibenolol

Yl

-2.00 -2.67 -36.65 0.46 0.14 -1.59 -3.18 -2.16 0.38 -1.02 -56.45 -121.50 -104.52 -27.73 -27.04 -28.49 0.03 0.23 -0.63 0.03 -6.66 0.45 0.43 -303.58 -52.26 -1.45 -0.1 1 0.44 -47.67 -1.22 -10.82 0.46 0.47 0.06 0.35 -1.78 -27.44 -41.23 -2.80 -0.03 0.47 -27.55

log IC 4.22 3.95 -3.43 1.86 2.29 -4.08 5.30 -3.22 0.01 -0.06 2.49 -7.07 -6.79 -5.56 -0.72 -1.47 3.93 0.17 0.8 1 4.49 1.33 2.22 0.3 1 -2.33 0.89 -1.03 0.68 -0.30 -3.33 6.61 1.55 4.79 4.06 5.38 2.51 5.31 -3.02 0.01 -2.72 2.29 2.94 -1.35

classification

False actives 7/40 = 17.5%; false inactives 8/42 = 19.05: overall accuracy 67/82 = 81.7%.

of the positions of the aromatic ring’s substituents (compare the values for the 0-, m-,and p-hydroxybenzoic acids). Likewise, said difference seemed to be more sensitive to the inductive effect of the ring’s substituents. Thus, p-methylpropiophenone, possessor of a weak positive inductor group (-CH3) shows a value of A4xp(0.050) much greater than that ofp-chloropropiophenone (0.016),which in the same position possesses a strong negative inductive group (-Cl). On the other hand, the difference A4xPseemed to be related more to the effect of aromaticity and the conjugation of electrons ?r and n, which conditions the value of the dipolar moment. Thus, the difference is greater for the 3,4-dimethylacetophenone than for the 2,4-dimethylacetophenonein concordance with the dipolar moments values, also being greater in the 2-(2-propenyl)phenol (where the ?r of the propenyl electrons are not conjugated with those of the aromatic ring) than in the 2 4 1-propenyl)phenol, where the ?r electrons are conjugated. Another fact concerns the analgesic character; the existence of negative inductive substituents (-Cl, -OH)in the “para” position of the aromatic ring hinder theanalgesic action (see the cases of 5-chlorosalicylic andp-hydroxybenzoic acids and of p-chloropropiophenone). Moreover, the existence of “meta” substituents, with respect to the carbonilicgroup, either hinder (for example in the case of 3,4-dimethylacetophenone)

or make difficult (3-chlorosalicylic acid) the analgesic activity. On the contrary, the existence of “ortho” substituents assure it. Since the presence of negative inductive groups in the para position implies a notable fall in the molecular dipole moments (see Table I ) , the influence of this physicochemical parameter on the analgesic character seems reasonable. Nevertheless, although the dipolar moment changes within each test group in a manner parallel to the differences (4xP- 4xvpo), it does not seem to be the determinant factor for analgesic character. act, inside each group, Moreover, the differences (4xp- 4xvp) as the discriminant. In fact, with only the exception of the phenolic derivatives, in all other cases such differences are higher for those molecules showing analgesic activity. However the foresaid differences are not useful by themselves as discriminant functions, given their high variability from one group to another. Thus, the analgesic character seems to be related to both the substituent’s position in the aromatic ring and the molecular dipole moments, which may, in turn, be related to the inductive and resonant effects. This is a reason why the cases of 2 4 1-propeny1)phenol and 2-(2-propenyl)phenol are significant; the first product possesses analgesic activity;” the second, showing only antiinflammatory activity,’* does not. This suggests that this combination effect must play an important role in the analgesic action. In fact,

1200 J . Chem. InJ Comput. Sci., Vol. 34, No. 5, 1994

GALVEZET AL.

Table 4. Results Obtained To Apply the Discriminant Functions (eqs 5 and 6) on a Group of 171 Molecules (29 with Analgesic Activity (+) and 142 Inactives (-)P molecule 2,4-acetophenone 2-amino-4-picoline 2 4 1-propeny1)phenol acemetacin alminoprofen amino-C 1-thenoxazole 2-amino-4-picoline diplosal acetate aniline antipyrine apazone bumadizon cinmetacin 1-(4-chlorophenyl)propanol chlorthenoxazine

Yl

log IC

0.43 0.41 0.24 0.34 0.11 0.33 0.14 0.38 0.45 0.37 0.31 0.46 -0.53 0.47 0.47

1.82 2.48 1.98 1.32 1.39 2.92 2.91 2.74 2.15 2.78 3.95 1.46 1.10 1.17 2.74

-0.20 -2.26 -0.40 0.45 -0.42 -0.62 0.25 -0.84 -1.36 -0.34 -2.23 -2.35 0.12 -0.10 -1.81 0.18 -2.94 -1.03 0.44 -1.28 -6.55 -0.68 -0.04 -0.25 -1.18 0.16 0.32 0.43 0.40 -6.81 -3.85 -9.49 -23.96 -0.27 0.45 -11.17 -0.36 -1.94 0.41 -1.04 -33.04 -14.41 -0.39 -27.79 -27.50 -38.62 -32.95 -27.96 -32.35 -88.01 0.27 -9.61 -1.69 -54.86 -35.91 -1.48 0.24 -0.98 -0.07

3.40 4.30 1.64 -0.10 -0.52 4.74 3.83 0.65 -0.68 -0.94 6.43 -0.51 0.21 -0.01 -0.90 0.68 3.54 1.09 0.63 -0.1 1 0.99 0.18 4.77 3.69 4.52 3.13 2.36 3-06 2.05 1.34 0.03 -0.33 -5.45 2.20 3.17 -0.24 -1.55 1.62 2.30 -2.00 -1.20 -7.81 -0.12 -0.56 -0.38 -0.94 -0.93 -1.30 -0.81 -6.71 2.11 7.31 1.93 -0.57 -0.51 -0.48 3.97 1.27 4.41

classification

molecule

Compounds with Analgesic Activity etersalate etodolac phenacetin fenbufen fenclofenac fenol ibufenac indoprofen morazone pirprofen methyl salicyclate sulfadiazine viminol zomepirac

+ + + + + + + + + + + + +

log IC

classification

-0.3 -27.76 -2.32 0.47 0.42 0.46 0.30 0.21 0.46 0.42 0.17 0.14 -0.30 0.47

2.46 1.69 2.43 0.37 1.52 2.26 1.52 1.17 2.54 1.17 2.62 1.18 0.24 1.61

+ + + + + + + + + +

-107.57 -0.19 0.47 0.44 0.20 -1.20 -0.23 -1 50.79 0.27 -1 -82 0.47 0.12 0.47 -1 1.69 -0.78 -0.09 0.39 -0.29 -1.03 0.32 0.45 -17.05 0.46 -22.95 -4.71 0.05 0.47 -15.04 0.33 0.32 -0.29 0.37 0.46 0.32 -109.13 -4.47 -29.32 0.36 -0.17 -4.35 -0.03 0.47 0.12 0.47 -5.13 0.46 -1.31 0.43 -15.68 -2.22 -9.35 0.42 0.46 0.28 -0.35 0.24 -0.37 -0.04 0.42

-5.28 1.31 1.34 -3.67 1.57 0.61 0.10 -2.67 1.15 2.99 3.18 1.78 -1.05 -8.08 3.76 3.19 2.51 2.60 2.74 1.65 2.79 -0.36 2.24 9.48 0.69 4.44 2.68 5.13 2.47 -2.54 2.53 1.17 2.57 1.77 -5.03 4.39 -4.06 1.77 2.37 s34 -0.87 2.49 2.80 0.71 4.59 -0.94 0.16 0.15 -0.35 4.73 2.10 -2.67 2.47 -0.14 6.21 -0.08 3.51 4.77 2.30

-

C:ompounds without Analgesic Activity (4-hydroxybuty1)guanine acicloguanosine almitrine astemizole azatadine brovincamine bucyclovir cetirizine cinnarizine cyproheptadine cytarabine clemastine chlorcyclizine chlorfenac chlorothen deptropine 6-desoxiaciclovir diphenhydramine diphenpyramide dimethindene doxylamine phenindamine ganciclovir 3HM-HBG 9 43-0hetoxy)guanine idoxuridine acebutolol acetic acid alprenolol amantadine amosulalol amotriphene arotinolol atenolol befunolol beclobrate bendrofluazide bevantolol bisoprolol bixin bopindolol brilliant blue FCF brontripelenamine bucumolol bufetolol bufur a1o1 bunitrolol bupranolol butofilolol canthaxanthin carazolol carminic acid carvedilol celiprolol cetamolol chlorothiazide citric acid clofibric acid clonitrate

-

+-

+-

+-

+ + -

-

-

+ -

-

-

+ -

-

-

-

cryptoxanthin curcumin epanolol epithiazide esmolol fenofibrate fenquizone flavoxanthin flumetramide glutamic acid glycerol indenolol indigo carmine C.I. acid green 5 malic acid maltol mephenesin mepindolol metipranolol metroprolol moprolol nadolol nadoxolol neohesperidin DHC nicofibrate nipradilol oxprenolol pentrinitrol pindolol ponceau 3R practolol pronethalol propionic acid propranolol rhodoxanthin riboflavine rubixanthin saccharin sorbic acid sorbitol sotalol styramate succinic acid sulfinalol tartaric acid tartrazine xipamide isothipendyl clomestrone eritadenine meglutol melinamide m-aminobenzoic acid mequitazine nebularine methaphenilene methyi gallate DHFG PABA

+ + -

+ + -

+

-

+

+-

+ + +

-

+

J. Chem. If. Comput. Sci., Vol. 34, No. 5, 1994 1201

TOWLOGICAL APPROACH TO ANALGESIA Table 4 (Continued) molecule

Yl

log IC

pyrathiazine ribavirin pyrilamine talastine thonzylamine trifluridine tripelennamide aryldone AzddBrU AzddOHU AzddOMeU AzddPropinyloxyU

0.06 -2.09 -1.53 0.17 -0.12 -0.40 -0.40 -1.84 0.46 0.09 0.04 0.27

-0.55 6.63 1 .05 0.91 2.12 4.68 1.04 -0.26 3.87 5.31 5.27 5.17

(I

classification molecule Compounds without Analgesic Activity AzddPropenyloxy U AzddCyanoMeOxyU AzddtiocyanoU AzddSMeU AzddC AzddFC AzddMeC Azdd3-( 3oxopropenyl)T Azdd6azatimidine ddC merocianine zidovudine

+

YI

log IC

classification

0.25 0.15 0.47 -0.38 0.44 0.00 0.28 0.43 0.17 0.11 -1.21 0.28

5.14 5.22 4.08 3.86 4.64 5.23 4.60 4.78 5.24 4.06 -0.8 1 4.79

-

-

-

-

Fakes active 6/29 = 20.7%;falses inactive 28/142 = 19.7%;overall accuracy 137/171 = 80.1%.

nearly all the nonnarcotic analgesics show a high relative percentage of the conjugated ?r and n electrons. Since Hammet's u function takes into consideration these effects, we thought of the possibility of introducing something like a topological Hammet function, which would include the aforementioned differences of the connectivity indices, in the form ut = a(4xp- 4XVP)

+ 6(4X,

- 4XV,)

+c

(1)

In this equation, ut represents a normalized value of the function, taking into consideration that the 0- and p hydroxybenzoic acids would have symmetrical values, that is to say, the same magnitude but with an opposite sign. The results for coefficients "a", "b",and "c" were a= 1

b=-1

~=0.04

As can be seen in Table 1 , although the utfunction, in various cases, takes on positive values for the molecules with analgesic activity and negative values for those with nonanalgesic activity, there are exceptions such as 3- and 5-chlorosalicylic acids or 4'-methyl- and 4'-chloropropiophenone. Moreover, the parameter ut does not act as a general discriminant function, demonstrated by the fact that multiple nonanalgesic compounds display very diverse utvalues (from -0.6 to +0.4). It is, therefore, necessary to find a general discriminant function for all the analgesics. Since the nonnarcotic analgesic action is related to the inhibition of cy~looxigenase,~3 we can use the obtained connectivity function with this property as a starting point for the discriminant analysis. log ICs0 = (-2.643~vp) + (1 .g4xP)- ( 1O.2g4xc)+ (1 .244x,)

N = 20

R = 0.8457 F = 9.42

p

+ 3.056 (2)

SE = 0.6281

< 0.0005

Where ICs0 = Inhibition Concentration-50 (pM) in bovine seminal vesicle13. As illustrated in Table 2, the majority of the analgesics shows theoretical log ICs0 values between 0 and 3, which is why it was chosen as the first discriminant function. Although many nonanalgesics possess values in this interval, few analgesics have them outside it, which is why this function can be considered as a necessary but not sufficient condition. With the objective of finding an efficient discriminant function, able to identify the nonanalgesic molecules, we selected an extensive group of compounds (40 with contrasted analgesic activity and 42 with no analgesic activity) and we

obtained the linear discriminant function using the connectivity indices up to the fourth order as independent variables. The chosen discriminant function was Y = (-1.325'~) + ( 4 . 6 7 ' ~ + ) (1 .96'xV)- (6.562x') (4.253xp)- (4.1 13xc) (2.6g3xVp) (13.3l3xVc)

+

+

+

(1 .2g4xP)+ (1 1 .754xc)+ ( I z 4 x , ) - 0.04 (3)

N = 82

F = 9.32

Ustatistics (Wilks' A) = 0.7976

If we use Yl = 0.47 - Y2, the value that Y1 takes on for a certain compound will indicate to us its analgesic character; thus, if Yl > 0, the compound will have theoretical analgesic activity, and if Y1 < 0, the compound will be inactive. In this manner, and working with both equations, a molecule will be selected as active if 3 > log ICs0 > 0 and Y1 > 0 or as inactive if Y1 C 0. Table 3 contains the classification results obtained after thediscriminant log ICs0 and Y1 functions with their respective discriminant functions were applied to each molecule. As can be observed, the average degree of success is superior to 80% (80.8%). The validity of the obtained discriminant function was confirmed when it was applied to a test group, meaning an extensive group of compounds not used in the discriminant function. Table 4 shows the results. The percent of success is similar to that obtained with the group used in the discrimination, which vouches for said discriminant functions. To extend the study, we widened the gt calculation for a group of analgesics whose common action mechanism is the inhibition of the cyclo~xygenase.~~ Since the molecule's form should have played an important part in the fixation of the enzyme, we used an E form factor, introduced by US,^ and which we define as E = S/L2

(4)

where "S"represents the "molecular surface parameter" and "L" the "topological molecular length", that is, the counted distance in the number of edges or links, between the molecule's two most separate atoms by the shortest means. S is calculated as the sum of the contributions for each molecular fragment, according to the values illustrated in Table 5 . It is important to point out that, only in relation to their contribution to the surface parameter, S , the double bonds are considered as simple. It is evident that, despite the simplicity of its calculation, the E index gives a measure of the molecular form; thus, molecules with high E values, like ASA or salicylic acid, show

1202 J . Chem. In$ Comput. Sci., Vol. 34, No. 5, 1994

GALVEZET AL.

Table 5. Contribution by Different Molecular Fragments to the Value of S 6 group /

-

A

contribution 2

+

0

6 12

M

20

Y

10

contribution

group

28

4

Y

n

0

14

36

18 49.5

I

I

I

I

..

-

.

.

JL

I

....

I.

I

0.m ot 2-(2-pmpnyl) phenol m h)dmx)benme acid

.

I.

0.40

Figure 1. Plot of potency, P,vs ut, for a wide group of analgesics, excepting aril-acetic derivatives.

24

a similar circular symmetry form, whereas those with a low value, such as tolmetin, show a form with greater excentricity. Table 2 illustrates the values for the potency, P (relative to ASA = l), ICs,, the E form factor, and ut for a wide group of analgesics, including compounds from different families: Aril-acetic, aril-propionic, and antranilic derivatives, as well as antipirines, salicylates, and oxicams. One point to be made is that there do not exist potent analgesics (P > 10) with values of E > 1.60. This result is logical since the union to the enzyme should be conditioned by the form, in such a way that the molecules close to a circular symmetry do not "fit" in the enzyme's union site. The form would be, therefore, a necessary but not sufficient condition (see the fenclofenac and alclofenac cases or the aminopirine case in which the low E value does not imply high potency). If we analyze the results by chemical families, we observe that the potency does not depend on that of the value of ut in the case of the aril-acetic derivatives, for those which, save for the two aforementioned exceptions (alcofenac and fenclofenac), the form seems to be decisive. Thus, all of them show values of E < 1.60 and high potencies, except etodolac with E = 1.81 and low potency. For the rest of the chemical families the analgesic potency is clearly influenced by ut. Figure 1 illustrates the representation of potency versus ut for the rest of the studied analgesics. As can be seen, two regions are clearly distinguished: the first, comprised between ut = 0 and ut = 0.8, for which no potent analgesics appear; the second, between ut = 0 and ut = -0.6 approximately, for which the potency increases with the absolute value of ut. Empirical equations which led to good adjustments were

derivatives), the parameter ut seems to have a discriminant action. Thus, for these kinds of compounds the ut values between 0 and -0.1 imply nule (or very low) analgesic activity (Figure 1). All the values between - 0 . 1 and +0.8 adjust reasonably well to eq 6. 4. CONCLUSIONS

The preceeding results clearly demonstrate that by adequate choice of topological indices it is possible to discriminate, with good efficiency, the minor (AINA) analgesic activity. Among these descriptors we may emphasize the connectivity ones, as well as the shape factor E. Moreover, it is also possible to predict the analgesic potency through the parameter ut,which is obtained as a linear combination of the differences (4xp4xvp)and (4xpc- 4xvpc).Those results illustrate the amazing ability of the topological descriptors in the rational discrimination of the analgesic character and, therefore, in the design of new drugs.16 Consequently, the rational design of new analgesics demands the following conditions: (A) Analgesic activity

(B) Potency (ASA = 1) (Bl) Aril acetic derivatives: only shape factor's influence; E < 1.60 implies high potency (P > 10) (B2) Other groups (B2.1) Phenyl derivatives: ut> 0, potency similar to ASA (P= 1); -0.1 < ut < 0, low potency (P< 0.5); -0.6 < ut < -0.1, high potency (P> 10) (B2.2) The rest: ut > 0, potency similar to ASA (P = 1); ut < 0, potency increases with -ut

(a) for -0.6 < ut < 0

P = K[1 - eaut]+ K'/( 1 + ey4)

ACKNOWLEDGMENT (5)

(b) for 0 < ut < 0.8

P = K'/(l

I

p-hydmr)benzmc 4M Z C I Benzotc sa

18

I

I

4.40

4.60

+ ey4)

The authors express their acknowledgment to CICYT, Grant SAF92-0684, (Spanish Ministery of Science) for its economical support of our research development. REFERENCES AND NOTES

(6)

with parameter values, K = 80, p = 2.35,K' = 2.5, and y = -1962. If the representation potency (P)versus ut is made for analgesics that only include one benzenic ring (Le. phenyl

(1) Rouvray, D. H.; El-Basil, S. J . Mol. Srrucr. 1988, 165, 9-20. (2) Kier, L. B.; Hall, L. H. Molecular Connecrluiry in Srructure-Actiuiry Analysis; Research Studies Press: Letchworth,England, 1986; pp 225-

246. (3)

Soler, R. M.; Garcfa, F.; Anton, G.; Garcfa, R.; Perez, F.; GAlvez, J. J . Chromatogr. 1992, 607, 91-95.

TOPOLOGICAL APPROACH TO ANALGESIA (4) Kupchik, E. J. Quant. Srrucr.-Acr. Relat. 1988, 1, 57-59. (5) Gblvez, J.; Garcfa, R.; Bernal, J.; Garcia, F. An. R. Acad. Farm. 1991, 57, 533-546. (6) Garcfa, R.; Gblvez, J.; Moliner. R.; Garcia, F. Drug Inuest. 1991,3 (9, 344-350. (7) Gblvez, J.; Garcia-Domenech, R.; Salabert, M. T.; Soler, R. J . Chem. Inf. Compur. Sci. 1994, 34, 52CL525. (8) Stewart, J. J. P. MOPAC. QCPE Bull. 1983, 3, 101. (9) Kier, L. B.; Hall, L. H. Molecular Connectiuity in Structure-Actiuiry Analysis; Research Studies Press: Letchworth, England, 1986;pp 252254. (10) Garcia, F.; Moliner, R.; Galvez, J.; Garcia, R. Rev. Farmacol. Clin. ExP., 1990, 7 (4), 209-211. (11) Gblvez, J.; Garcfa, R.; Moliner, R.; Garcia, F. Spanish Patent No. P9101034, 1991.

J. Chem. InJ Comput. Sci., Vol. 34, No. 5, 1994 1203 (12) Chang, M. N.; Jensen, N. P.; Hammond, L.; et al. US patent number 4.778.818, 1988. (13) Garcia, F. Disefio por topologfa molecular, slntesis y ensayos farmacolbgicos de nuevos fhrmacos analgbim. Doctoral Thesis, Faculty of Pharmacy, University of Valencia, Spain, 1991; p 67. (14) The Merck I d e x , 4.Mer& ; & c o . , Inc.: Rahway, NJ, 1989, THER-6. (15) Florez, J. Human Pharmacology; Edicione-s Cientfficas y Tknicas, S. A.: Barcelona, Spain, 1992; pp 334-335. (16) Gblvez, J.; Garcla, R.; Salabert, T.; et al. New Drugs Designed by Molecular Topology (FTZl), Supplement to Pharmaceutisch Weekblad Scientific Edition, October 1992; Vol. 14, No. 5, p 32 (Abstracts of the First European Congress of Pharmaceutical Sciences) Amsterdam, The Netherlands.